View All Items
- Title
- A FRAMEWORK FOR EFFICIENT DATA DISTRIBUTION IN PEER-TO-PEER NETWORKS.
- Creator
-
Purandare, Darshan, Guha, Ratan, University of Central Florida
- Abstract / Description
-
Peer to Peer (P2P) models are based on user altruism, wherein a user shares its content with other users in the pool and it also has an interest in the content of the other nodes. Most P2P systems in their current form are not fair in terms of the content served by a peer and the service obtained from swarm. Most systems suffer from free rider's problem where many high uplink capacity peers contribute much more than they should while many others get a free ride for downloading the content...
Show morePeer to Peer (P2P) models are based on user altruism, wherein a user shares its content with other users in the pool and it also has an interest in the content of the other nodes. Most P2P systems in their current form are not fair in terms of the content served by a peer and the service obtained from swarm. Most systems suffer from free rider's problem where many high uplink capacity peers contribute much more than they should while many others get a free ride for downloading the content. This leaves high capacity nodes with very little or no motivation to contribute. Many times such resourceful nodes exit the swarm or don't even participate. The whole scenario is unfavorable and disappointing for P2P networks in general, where participation is a must and a very important feature. As the number of users increases in the swarm, the swarm becomes robust and scalable. Other important issues in the present day P2P system are below optimal Quality of Service (QoS) in terms of download time, end-to-end latency and jitter rate, uplink utilization, excessive cross ISP traffic, security and cheating threats etc. These current day problems in P2P networks serve as a motivation for present work. To this end, we present an efficient data distribution framework in Peer-to-Peer (P2P) networks for media streaming and file sharing domain. The experiments with our model, an alliance based peering scheme for media streaming, show that such a scheme distributes data to the swarm members in a near-optimal way. Alliances are small groups of nodes that share data and other vital information for symbiotic association. We show that alliance formation is a loosely coupled and an effective way to organize the peers and our model maps to a small world network, which form efficient overlay structures and are robust to network perturbations such as churn. We present a comparative simulation based study of our model with CoolStreaming/DONet (a popular model) and present a quantitative performance evaluation. Simulation results show that our model scales well under varying workloads and conditions, delivers near optimal levels of QoS, reduces cross ISP traffic considerably and for most cases, performs at par or even better than Cool-Streaming/DONet. In the next phase of our work, we focussed on BitTorrent P2P model as it the most widely used file sharing protocol. Many studies in academia and industry have shown that though BitTorrent scales very well but is far from optimal in terms of fairness to end users, download time and uplink utilization. Furthermore, random peering and data distribution in such model lead to suboptimal performance. Lately, new breed of BitTorrent clients like BitTyrant have shown successful strategic attacks against BitTorrent. Strategic peers configure the BitTorrent client software such that for very less or no contribution, they can obtain good download speeds. Such strategic nodes exploit the altruism in the swarm and consume resources at the expense of other honest nodes and create an unfair swarm. More unfairness is generated in the swarm with the presence of heterogeneous bandwidth nodes. We investigate and propose a new token-based anti-strategic policy that could be used in BitTorrent to minimize the free-riding by strategic clients. We also proposed other policies against strategic attacks that include using a smart tracker that denies the request of strategic clients for peer listmultiple times, and black listing the non-behaving nodes that do not follow the protocol policies. These policies help to stop the strategic behavior of peers to a large extent and improve overall system performance. We also quantify and validate the benefits of using bandwidth peer matching policy. Our simulations results show that with the above proposed changes, uplink utilization and mean download time in BitTorrent network improves considerably. It leaves strategic clients with little or no incentive to behave greedily. This reduces free riding and creates fairer swarm with very little computational overhead. Finally, we show that our model is self healing model where user behavior changes from selfish to altruistic in the presence of the aforementioned policies.
Show less - Date Issued
- 2008
- Identifier
- CFE0002260, ucf:47864
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0002260
- Title
- STUDENT DIGITAL PIRACY IN THE FLORIDA STATE UNIVERSITY SYSTEM:AN EXPLORATORY STUDY ON ITS INFRASTRUCTURAL EFFECTS.
- Creator
-
Reiss, Jeffrey, Cintron, Rosa, University of Central Florida
- Abstract / Description
-
Digital piracy is a problem that may never disappear from society. Through readily available resources such as those found in a university, students will always have access to illegal goods. While piracy is a global phenomenon, an institutionÃÂ's resources combined with the typical college studentÃÂ's lack of funds makes it more lucrative. Students use a number of methods to justify their actions ranging from previewing media to bringing justice...
Show moreDigital piracy is a problem that may never disappear from society. Through readily available resources such as those found in a university, students will always have access to illegal goods. While piracy is a global phenomenon, an institutionÃÂ's resources combined with the typical college studentÃÂ's lack of funds makes it more lucrative. Students use a number of methods to justify their actions ranging from previewing media to bringing justice to a corrupt company. While trying to understand the mindset of pirates is one route to deal with piracy, corporations attempted to alleviate the situation using added software encoding. These messages are not always effective, and in some cases caused further damage to consumer morale. Furthermore, students such as Joel Tenenbaum, who continued to pirate music despite warnings from his parents and the recording industry, exemplify the type of person that is unfazed by legal threats, leading to a question of ethics. Students may not feel that downloading is stealing despite numerous warnings from the Digital Millennium Copyright Act and other major media organizations. The predominant solution used by universities involves monitoring the studentsÃÂ' network connection to detect Peer-to-Peer (P2P) connections or other connections that involve the transferring of copyrighted goods. Unfortunately, the current tools contain flaws that a crafty student may easily circumvent, undermining any attempts a universityÃÂ's IT department may use to deter piracy. This study explored the nature of piracy prevention tools used by IT departments in the Florida State University System in order to determine their relative effectiveness. The study also looked into the opinions of the Information Security Officer in terms of alternative piracy prevention techniques that do not involve legal action and monitoring. It was found that most institutions do not use a formal piece of software that monitors for infringing data. They also stated that while their current techniques can do its required task, it was not perfected to a point where it could run autonomously. Furthermore, institutions agreed that students lack proper ethics and concern over the matter of copyright, but were not fully convinced that other preventions methods would be effective. The study ultimately considered monitoring techniques a short-term solution and that more research should be put into finding long-term solutions. It also implied that IT departments should be better funded in order to keep up with the technological gap.
Show less - Date Issued
- 2010
- Identifier
- CFE0003131, ucf:48645
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0003131